You are on page 1of 5

International Journal of Computer Techniques - Volume 2 Issue 1, 2015

RESEARCH ARTICLE

OPEN ACCESS

Image Fusion Based on Medical Images Using DWT and


PCA Methods
Mr. Roshan P. Helonde1, Prof. M.R. Joshi2
1

M.E. Student,
Department of Computer Science and Information Technology,
H.V.P.Ms College of Engineering & Technology, Amravati, Maharashtra, India.
2
Assistant Professor,
Department of Information Technology,
H.V.P.Ms College of Engineering & Technology, Amravati, Maharashtra, India
--------------------------------------------------************************************------------------------------------------

Abstract:
Image fusion is the process of combining relevant information from a set of images into a single image in
which the fused image contains more information than any of the input images. This technique improves the quality
of data. Image fusion is one of the important re-processing steps in digital image reconstruction. Now-a-days,
medical image fusion is one of the upcoming fields which helps in easy diagnostics and helps to bring down the time
gap between the diagnosis of the disease and the treatment. In Magnetic Resonance Image (MRI), anatomy and soft
tissues are visible and it has high spatial resolution. In Computed Tomography (CT) images bony structures appears
brighter. Analysis is done to determine the image fusion algorithm which is more suitable for clinical diagnosis.
Analysis is also done on image quality assessment parameters of image fusion. This paper presents image fusion
techniques and image quality assessment parameters such as primitive fusion (Averaging Method, Select Maximum,
and Select Minimum), Discrete Wavelet transform based fusion, Principal component analysis (PCA) based fusion
etc. Comparison of all the techniques concludes the better approach for its future research.
Keywords : - Image Fusion, Discrete Wavelet Transform (DWT), Principal Component Analysis (PCA),
Mean Square Error (MSE), Peak signal to noise ratio (PSNR).
--------------------------------------------------************************************------------------------------------------

INTRODUCTION :
Image Fusion is a process of combining
the relevant information from a set of
images, into a single image, such that the
resultant fused image will be more
informative and complete than any of the
input images. For example, CT and MR
images may be fused as an aid to medical
diagnosis. MR and visible images may be
fused as an aid to pilots landing in poor
weather condition or microwave and visible
images may be fused to detect weapons. The
ISSN: 2394-2231

fusion process must satisfy the following


requirements such as it should preserve all
relevant information in the fused image,
should reduce noise and should minimize
any artifacts in the fused image. There are
two approaches to image fusion, namely
Direct Fusion and Multi resolution fusion. In
Direct fusion, the pixel values from the
source images are summed up and taken
average to form the pixel of the composite
image at that location. Multi resolution
fusion uses pyramid or wavelet transform

http://www.ijctjournal.org

Page 75

International Journal of Computer Techniques - Volume 2 Issue 1, 2015

for representing the source image at multi


scale. Because of the characteristic of image
fusion technology, it has recently been
widely applied in various fields such as auto
target recognition, computer vision,
Concealed weapon detection and medical
image processing etc.. The advantages of
image fusion are improving reliability and
capability.
I.

FUSION TECHNIQUES
A. Discrete Wavelet Transform (DWT)

The Discrete Wavelet Transform (DWT)


[2][5] also converts the image from the
spatial domain to frequency domain. The
image is divided
ded by vertical and horizontal
lines and represents the first-order
order of DWT,
and the image can be separated with four
parts
rts those are LL1, LH1, HL1 and HH1. In
additional, those four parts are repr
represented
four frequency areas in the image. For the
low- frequency
uency domain LL1 is sensitively
with human eyes. In the frequency domains
LH1, HL1 and HH1 have more detail
Information more than frequency domain
LL1.

Fig1:
1: Frequency distribution of DWT

1)

Algorithms of DWT

ISSN: 2394-2231

Wavelet transform is first performed on


each source images to generate a fusion
decision
on map based on a set of fusion rules.
The fused wavelet coefficient map can be
constructed from the wavelet coefficients
[7][9]of
of the source images according to the
fusion decision map. Finally the fused image
is obtained
btained by performing the inverse
wavelet transform.
The algorithm of image fusion using
DWT [8] has following common steps
applicable to proposed methods of fusion.
a) Accept the two input images.
b) Resize both the images to 256 x 256.
c) Convert to Gray scale image.
d) Convert to double precision format.
e) Take Discrete Wavelet Transform of both
the images.
f) Let for first image OUT bands be HHa,
HLa, LHa, LLa and for second image be
HHb, HLb, LHb, LLb.
B. Principal Component Analysis (PCA):

It is a mathematical tool from applied


linear algebra .It is a simple non-parametric
non
method of extracting relevant information
from confusing data sets. PCA [3][6] is a
useful statistical technique that has found
application in fields such as face recognition
recognit
and image compression, and is a common
technique for finding patterns in data of high
dimension. The origins of PCA
P
lie in
multivariate data analysis, it has a wide
range of other applications
ns PCA has been
called, one of the most important results
from
m applied linear algebra and perhaps
perhap its
most common use is as the first step in
trying to analyses large data sets. In general
terms, PCA uses a vector space transform to
reduce the dimensionality of large data sets.

http://www.ijctjournal.org

Page 76

International Journal of Computer Techniques - Volume 2 Issue 1, 2015

Using mathematical projection, the ori


original
data set, which may have involved many
variables, can often be interpreted in just a
few variables (the principal components).
The normalized components PI and P2 are
computed from the obtained eigenvector.
The fused image is given by equation,
If(x,y)=P1I1(x,y)+P2I2(x,y)

Fig 2: Image Fusion Process using PCA

1.

Algorithm of PCA:

Let the source images be arranged in


two-column
column vectors. The steps followed to
project this data into 2-D
D subspaces are:
1. From the input images matrices arrange
the data into column vectors. The resulting
matrix Z is of dimension 2xn.
2. Then Compute the empirical mean along
each column. The empirical mean vector Me
has a dimension of 1 x 2.
3. Subtracting the empirical mean vector Me
from each column of the data matrix S. The
resulting matrix X is of dimension 2 x n.
4. Find the covariance matrix C of X i.e.
C=XXT mean of expectation = cov(X)
cov(X).
5. Compute the eigenvectors V and
eigenvalue D of C and sort them by
decreasing eigenvalue. Both V and D are of
dimension 2 x 2.

ISSN: 2394-2231

6. Finally consider the first column of V


which corresponds to larger eigenvalue to
compute PI and P2 as,

II.

MRI AND CT ANALYSIS

The paper forms fused images of CT


[1][5] and MRI [1] image. Fused images can
be created
d by combining information from
multiple modalities, such as Magnetic
Resonance Image (MRI) and Computed
Tomography (CT).CT
T images are used more
often to as certain differences in tissue
density depending upon their ability to block
X-rays while MRI [5][11] provides good
contrast between the different
ferent soft tissues of
the body, which make it especially useful in
detecting brain tissues, and
an cancers. The
fused image from multiple images produce
an image which contains combined
complementary and redundant information
informat
provided by both the source images i.e. the
size of the tumor, the location through the
various pixel values of the gray scale
images,
es, hence resulting into better visibility
of tumor.
III.

PERFORMANCE METRICS

Image fusion performance can be


divided into two categories one with and one
without reference
eference images. In the present
work, the performance measures are used to
evaluate the performance of various fusion
methods such as Peak signal to noise Ratio
[6], Mean Square Error, Average Difference
[2], Minimum Difference [12], Maximum
Difference
and
Normalized
Cross
Correlation Methods. MR and CT images

http://www.ijctjournal.org

Page 77

International Journal of Computer Techniques - Volume 2 Issue 1, 2015

[1] are taken as the reference image in the


calculation of performance metric values.

than other numbers. In this context, suppose


that we measure the quality of t, as a

A. Peak signal to noise Ratio

measure of the center of the distribution, in

The PSNR is most commonly used as a

terms of the mean square error [8].

measure of quality of reconstruction of loss


compression

codecs

(e.g.,

for

image

compression). The signal in this case is the

by

compression.

%"# !"#

, ]

Average Difference (AD) [7]is defined

When

comparing compression codecs it is used as

C. Average Difference

original data, and the noise is the error


introduced

as:

an approximation to human perception of


reconstruction quality, therefore in some
cases one reconstruction may appear to be
closer to the original than another, even
though it has a lower PSNR [12] (a higher
PSNR would normally indicate that the
reconstruction is of higher quality).

D. Minimum Difference

The PSNR is calculated by using following


formula [10].
= 10

The large value of AD means that the


pixel values in the reconstructed image are
more deviated from actual pixel value
[10][3]. Larger value of AD indicates image
is of poor quality.

10

After multiplying PSNR by 10 log 10 PSNR


is converted into decibel i.e db is a unit of
PSNR.

C = normxcorr2 (template, A) computes


the normalized cross-correlation of the
matrices template and A. The matrix A must
be larger than the matrix template for the
normalization to be meaningful. The values
of template cannot all be the same. The
resulting matrix C contains the correlation
coefficients, which can range in value from 1.0 to 1.0.
E. Maximum Difference

B. Mean Square Error

In a sense, any measure of the center of a


distribution should be associated with some
measure of error. If we say that the number t
is

good

measure

of

center,

Difference between any two pixels such


that the larger pixel appears after the
smallest pixel. The large value of maximum
difference [9] means that image is poor in
quality.

then

presumably we are saying that t represents


the entire distribution better, in some way,
ISSN: 2394-2231

http://www.ijctjournal.org

Page 78

International Journal of Computer Techniques - Volume 2 Issue 1, 2015


F. Normalized Cross Correlation

The Normalized Cross Correlation


[12][7] between the reference image R and
the fused image F is defined as:

[6]

[7]

[8]

[9]

IV.

CONCLUSIONS AND FUTURE WORK

Finally this paper concludes that a image


fusion algorithm based on combination of
DWT and PCA with morphological
processing will improve the image fusion
quality and may be the future trend of
research regarding image fusion. In this
paper we study the Image Fusion techniques
such as Discrete Wavelet Transform (DWT)
and Principal Component Analysis (PCA).
In future we discuss on the result occurring
by using these techniques experientially and
compare those values to check which
technique is giving correct result. Thus the
two different modality images are fused
using the various fusion rules based on the
Discrete Wavelet Transform.

[10]

[11]

[12]

Intelligent Computing Techniques (ICICT), IEEE 2014, pp.


765-770.
Manjusha, D. Udhav, B., Image Fusion and Image Quality
Assessment of Fused Images International Journal of Image
Processing, 2009, 4(5), pp. 484-508.
Rajeswara Mahidar P, N.V.G.Prasad, V.L.N sastry , A
Novel Noisy Image Fusion Technique Using Dual-Tree
Complex Wavelets Transform; IJEST, Vol. 3 No. 6, June
2011.
Mianshu Chen, Ping Fu, Yuan Sun, Hui Zhang, Image
Retrieval Based on Multi-Feature Similarity Score Fusion
Using Genetic Algorithm, IEEE 2010.
De-gan Zhang, Chao Li, Dong Wang, Xiao-li Zhang, A
New Method of Image Data Fusion Based on FNN, IEEE
2010.
Naga Sravanthi Kota and G.Umamaheswara Reddy Fusion
Based Gaussian noise Removal in the Images using
Curvelets and Wavelets with Gaussian Filter, International
Journal of Image Processing, Vol. 5, No. 4, pp. 456-468,
October 2011.
Wu, K .Wang , C .and Li , L ., Image Fusion at Pixel Level
Algorithm Is Introduced and the Evaluation criteria ,
International Conference on Educational and Network
Technology (ICENT 2010) , pp.585-588.
De-gan Zhang, Chao Li, Dong Wang, Xiao-li Zhang, A
New Method of Image Data Fusion Based on FNN, IEEE
2010.

REFERENCES
[1]

[2]

[3]

[4]

[5]

Rui Shen, Irene Cheng, and Anup Basu, Cross-Scale


Coefficient Selection for Volumetric Medical Image
Fusion, IEEE Transactions On Biomedical Engineering,
VOL. 60, NO. 4, APRIL 2013, pp. 1069-1079.
Marcio H. Ribeiro Sales, Carlos M. Souza, Jr., and Phaedon
C. Kyriakidis, Fusion of MODIS Images Using Kriging
With External Drift, IEEE Transactions On Geoscience
And Remote Sensing, VOL. 51, NO. 4, APRIL 2013, pp.
2250-2259.
Chaunte W. Lacewell, Mohamed Gebril, Ruben Buaba and
Abdollah Homaifar, Optimization of Image Fusion Using
Genetic Algorithms and Discrete Wavelet Transform,
IEEE 2010, pp. 116-121.
M. Gonzlez-Audcana, J. L. Saleta, R. G. Cataln, and R.
Garcia, Fusion of multispectral and panchromatic images
using improved IHS and PCA mergers based on wavelet
decomposition, IEEE Trans. Geosci. Remote Sens., vol.
42, no. 6, pp. 12911299, Jun. 2004.
Om Prakash, Arvind Kumar and Ashish Khare, Pixel-level
image fusion scheme based on steerable pyramid wavelet
transform using absolute maximum selection fusion rule,
International Conference on Issues and Challenges in

ISSN: 2394-2231

http://www.ijctjournal.org

Page 79

You might also like